Molecular evidence and clinical importance of β‐arrestins expression in patients with acromegaly

Abstract β‐arrestins seem to have a role in endocytosis and desensitization of somatostatin receptor subtype 2 (sst2) and could be associated with the responsiveness to somatostatin receptor ligands (SRL) in patients with acromegaly. To investigate the in vivo correlation between β‐arrestins 1 and 2 with sst2, sst5 and dopamine receptor subtype 2 (D2) expressions, and the association of β‐arrestins with response to first‐generation SRL and invasiveness in somatotropinomas. β‐arrestins 1 and 2, sst2, sst5 and D2 mRNA expressions were evaluated by quantitative real‐time RT‐PCR on tumoral tissue of 96 patients. Moreover, sst2 and sst5 protein expressions were also evaluated in 40 somatotropinomas by immunohistochemistry. Response to SRL, defined as GH <1 μg/l and normal IGF‐I levels, was assessed in 40 patients. The Knosp‐Steiner criteria were used to define invasiveness. Median β‐arrestin 1, β‐arrestin 2, sst2, sst5 and D2 mRNA copy numbers were 478; 9375; 731; 156; and 3989, respectively. There was a positive correlation between β‐arrestins 1 and 2 (R = 0.444, P < 0.001). However, no correlation between β‐arrestins and sst2, sst5 (mRNA and protein levels) or D2 was found. No association was found between β‐arrestins expression and SRL responsiveness or tumour invasiveness. Although previous data suggest a putative correlation between β‐arrestins and sst2, our data clearly indicated that no association existed between β‐arrestins and sst2, sst5 or D2 expression, nor with response to SRL or tumour invasiveness. Therefore, further studies are required to clarify whether β‐arrestins have a role in the response to treatment with SRL in acromegaly.


Introduction
Acromegaly is a chronic disease due to hypersecretion of growth hormone (GH), commonly caused by a GH-secreting pituitary adenoma (somatotropinoma). Surgery is the first-line treatment; however, approximately half of the patients will not be cured, especially those with macroadenomas. In such cases, medical adjuvant treatment is often necessary, and the first-generation somatostatin receptor ligands (SRL) are the first choice [1].
The first-generation SRL, octreotide and lanreotide, act mainly through somatostatin receptor subtype 2 (sst2), and the expression of this receptor is used as a predictor of response to SRL [2,3]. However, there are patients who, despite high sst2 expression levels, do not respond to SRL therapy [4]. Several studies have been conducted with the aim of identifying the molecular basis of partial or total resistance to these drugs, varying from reduced density of the sst subtypes [2], receptors mutation [5] to changes in signal transduction and cytoskeleton proteins involvement, etc. [6]. The recognition of a subset of patients unlikely to respond to SRL therapy might allow the rational use of these drugs in patients with acromegaly [7].
In this sense, it was recently described that the expression of barrestins could lead to changes in the signalling pathway of sst2 in somatotropinomas and consequently could be related to the responsiveness to first-generation SRL [6,8]. Specifically, b-arrestins bind to G protein-coupled receptors (GPCRs) after their phosphorylation by the G protein-coupled receptor kinase (GRK) type 2, allowing the recruitment of proteins, regulating the trafficking of proteins and the endocytic machinery. After stabilization of binding with GPCR, barrestins undergo conformational change that allows their connection with the endocytic machinery dependent on clathrin [9], allowing the internalization of b-arrestins-GPCR complex, leading to receptor desensitization [9]. The stability of the arrestin-receptor complex appears to be determinant for the recycling rate and internalization of GPCRs. Indeed, this process of arrestin-induced endocytosis of the sst2 was observed in cell cultures [10].
b-arrestins are considered multifunctional adaptor molecules, with several binding partners. In vitro study showed that b-arrestins regulate the activation of multiple components of MAPK cascade, such as ERK, redirecting these components from the nucleus to the cytoplasm and resulting in a decrease in proliferation [11]. In fact, changes in barrestins expression (at the mRNA and protein level) have been associated with more aggressive cancer phenotypes [12][13][14].
Based on the data reported to date, the aim of this study was to further investigate the putative in vivo correlation between the expression of b-arrestins 1 and 2 with sst2, sst5 or dopamine receptor type 2 (D2) and also the association between b-arrestins expression with the response to first-generation SRL and with the presence of cavernous sinus invasion in somatotropinomas.

Subjects and methods
This study was approved by the Ethics Committee of Hospital Universit ario Clementino Fraga Filho and Medical School/Universidade Federal do Rio de Janeiro. All patients signed an informed consent before entering the study.

Patients and tumours
Consecutive patients with acromegaly who had fresh tumour samples available obtained during the transsphenoidal surgery performed at the Instituto Estadual do C erebro Paulo Niemeyer and at the Hospital Universit ario Clementino Fraga Filho were included in this study. The tumour was immediately snap-frozen and stored at À80°C for molecular biology studies, and available formalin fixed paraffin-embedded tissues were used for immunohistochemistry.
The diagnosis of acromegaly was performed according to current criteria [1]. The maximum diameter of the tumour and the presence or absence of cavernous sinus invasion, using the modified Knosp-Steiner criteria [15], were evaluated by the same radiologist analysing the preoperative magnetic resonance image (MRI).

Criteria for cure and response to first-generation SRL
Patients with nadir GH levels after oral glucose tolerance test (OGTT) higher than 1.0 ng/ml or age-matched plasma high IGF-I levels three months after surgery were considered not cured. Subsequently, the biochemical response to medical treatment was evaluated by GH and IGF-I levels after 6 months of treatment with octreotide LAR at maximum dose of 30 mg or lanreotide autogel at maximum dose of 120 mg. Uncontrolled patients were those who had GH levels >1.0 lg/l and/or IGF-I levels higher than age-matched normal patients.

Methods
Hormonal assessment GH was assayed by a chemiluminescence (IMMULITE 2000; DPC -Diagnostic Products Corp., Inc., Los Angeles, CA, USA). The intra-and interassay coefficients of variation (CV) were 5.8 and 6.0%, respectively. The International Reference Preparation (IRP) for GH was the 98/ 574. IGF-I was measured by a chemiluminescence (IMMULITE 2000; DPC). The intra-and interassay CV were 3.6 and 6.6%, respectively. The IRP for IGF-I was the 87/518. The IGF-I results were expressed as percentage of the upper limit of normal range (%ULNR).

Immunohistochemistry
Paraffin-embedded tissue sections from somatotropinomas (3 lm) were dewaxed, rehydrated and submitted to heat-mediated antigen retrieval for sst2 and sst5 with Declere â (Cell Marque, Rocklin, CA, USA), pH 6.0 in a cooker pressure for 15 min. Endogenous peroxidase was inhibited in peroxide block (Cell Marque) for 10 min., followed by blocking unspecific immunoglobulin binding with background block (Cell Marque) for 10 min. before primary antibody incubation.
Rabbit monoclonal antibodies directed against sst2 antigen (UMB-1,  Marque) and revealed by the chromogen substrate diaminobenzidine (Liquid DAB, Cell Marque). The tissues were counterstained with haematoxylin. Normal human pituitaries obtained from autopsies were used as positive control.

Quantitative PCR
The expression of b-arrestins 1 and 2, sst2, sst5 and D2 in somatotropinomas was analysed by quantitative real-time RT-PCR (qPCR) using Sybergreen â method as previously described [18]. The ribonucleic acid (RNA) was isolated using Allprep â Universal kit (Qiagen, Hilden, Germany) and treated with DNase (Turbo DNA-free TM Kit, Invitrogen, CA, USA), according to the manufacturers' protocol. The quality and amount of the extracted RNA were evaluated using NanoDrop Lite Spectrophotometer (Thermo Fischer, Wilmington, DE, USA) and Qubit 3.0 Fluorometer (Life Technologies, Foster City, CA, USA), respectively.
The reverse transcription (RT) reaction was performed with 1 lg of total RNA in a final volume of 40 ll containing 109 RT Random Primers, 259 dNTP Mix, 50 U/ll MultiScribeTM Reverse Transcriptase, 20U/ll RNase Inhibitor and 109 RT Buffer [using the first strand cDNA synthesis kit (High-capacity cDNA reverse transcription kit, Life Technologies)]. The cycling condition was 25°C for 10 min., 37°C for 120 min. and 85°C for 5 min. GeNorm 3.3 Visual basic application for Microsoft Excel (http:// medgen.ugent.be/wjvdesomp/genorm/) was used to evaluate the best housekeeping genes that could be suitable as internal controls in the present cohort, as previously validated by Vandesompele et al. [19]. Specifically, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), bactin (ACTB) and hypoxanthine-guanine phosphoribosyl transferase (HPRT) were found to be stable genes using the qPCR analysis in our cohort of samples.
One lL aliquots of cDNA were amplified by qPCR using the primers specified in Table 1 and the Brilliant III SYBR Green Master Mix in the Stratagene Mx3000p instrument (Agilent, Madrid, Spain). Thermal cycling profile consisted of a pre-incubation step at 95°C for 3 min., followed by 40 cycles of denaturation (95°C for 20 sec.), annealing and extension (60°C for 20 sec.), and finally, a dissociation cycle to verify that only one product was amplified.
Standard curves were constructed using aliquots of purified PCR products serial diluted to obtain standards containing 10 0 , 10 1 , 10 2 , 10 3 , 10 4 , 10 5 and 10 6 copies of each transcript analysed per microlitre. As previously described [20], each standard curve was run in parallel with the samples to estimate the mRNA copy number in each sample. The geometric mean of the copy numbers for the three endogenous controls was calculated generating a normalization factor (NF) using the GeNorm 3.3 Visual basic application. Results were then reported as target gene copy number/NF.

Statistical analysis
Analyses were performed with SPSS version 23.0 for Mac (SPSS, Inc., Chicago, IL, USA). In the descriptive analysis, categorical variables were expressed through their percentage and frequency and numerical variables were described as median values (minimum-maximum). To compare numeric variables between groups, the Mann-Whitney nonparametric test was used. To correlate numeric variables, the Spearman test was used. P value <0.05 was considered significant.

Patient/sample characteristics
Of 96 patients who participated in the study, 44 patients were female (46%). The median age at diagnosis was 43 years (15-75 years). Data of tumour size were available from 72 patients (61 macroadenomas 9 11 microadenomas) and of tumour invasiveness from 33 patients (19 non-invasive 9 14 invasive). All invasive tumours were Table 1 Primer sequences, product sizes and GenBank accession numbers used for quantitative assessment of b-arrestin 1, b-arrestin 2, sst2, sst5 and D2 and housekeeping genes (ACTB, HPRT1, GAPDH) by real-time RT-PCR macroadenomas. The median GH at diagnosis was 18.8 lg/l (1.1-120), and IGF-I at diagnosis was 325%ULNR (101-734; Table 2). None of the patients received radiotherapy before surgery. Only 40 tumour specimens (from the 96 patients included in the study) were available for analysis of sst2 and sst5 by immunohistochemistry.
In relation to treatment with first-generation SRL before surgery, 21 patients did not have available data, 62 patients were treatment na € ıve and 13 patients were treated before surgery. Of the latter, two patients also used cabergoline.
Data regarding response to first-generation SRL were available in 40 and from these 23 (57.5%) were controlled. Of these, one patient was excluded of the sst2 and sst5 analyses because we did not have qPCR data available (probably due to the poor quality of the sample obtained).
The median sst2 mRNA levels was lower in patients who used SRL before surgery (n = 13) compared to those who did not (n = 62) [96 (12-23,954) 9 937 (34-13,995), respectively, P = 0.001], but we did not find difference in relation to b-arrestin 1 and 2, sst5 or D2 mRNA expression in patients who were treated or not before surgery.
Correlation between b-arrestins 1 and 2 with, sst2, sst5, D2 mRNA expression levels and sst2, sst5 protein levels We found a clear positive correlation between b-arrestins 1 and 2 (R = 0.444, P < 0.001). However, no correlation was found between b-arrestins 1 or 2 and sst2, sst5 or D2 mRNA levels or with the protein levels of and sst2 or sst5.
There was no significant correlation between b-arrestin 1 and barrestin 2 with sst2, sst5 and D2 mRNA levels in patients who were treated before surgery or in those treatment na € ıve. Given that sst2 mRNA levels were found to be lower in patients who used SRL before surgery compared to those who did not, we implemented a separated analysis in the case of sst2. Specifically, fifteen of the 40 patients with available data regarding the response to first-generation SRL treatment were excluded from this analysis because they did not have available data on the use of SRL before surgery (n = 5) or in the sst2 mRNA levels (n = 1), or because they  Expression levels are normalized against three housekeeping genes (glyceraldehyde-3-phosphate dehydrogenase-GAPDH, b-actin-ACTB and hypoxanthine-guanine phosphoribosyl transferase-HPRT). sst: somatostatin receptor, D2: dopamine receptor. had been pre-treated with SRL before surgery (n = 9). Of the available 25 patients, 13 (52%) were controlled. In these patients, the sst2 mRNA expression was also higher in patients who achieved biochemical control versus. uncontrolled patients [1204 (254-23,954) versus 478 (12-13,747), P = 0.04]. On the other hand, b-arrestins mRNA levels were similar between controlled and uncontrolled patients.

Discussion
b-arrestins are multifunctional adaptor molecules and they have several binding partners. They bind to various GPCRs and mediate their internalization [9], having a role in endocytosis and desensitization of sst2 [10]. Moreover, b-arrestins seem to be involved in regulation of cell proliferation [12]. Therefore, based on this information, it has been previously suggested that b-arrestins might be associated with the responsiveness to first-generation SRL in patients with acromegaly [21]. In the present study, we correlated the expression levels of barrestins with sst2, sst5 (at the mRNA and protein levels) or D2 expression and investigated the association of b-arrestins expression with SRL responsiveness and tumour invasiveness in somatotropinomas. We found low b-arrestin 1 mRNA levels compared to the expression of b-arrestin 2 in somatotropinomas, which is in agreement with previous studies [22][23][24]. Specifically, these low expression levels were also observed in normal rat pituitary tissue and in the rat GH3 cell line [25]. In this sense, our study indicates that the use of first-generation SRL before surgery did not influence b-arrestins 1 or 2 mRNA expression in agreement with the data reported by Gatto et al. [24].
Interestingly, we found clear positive correlation between the expression levels of b-arrestins 1 and 2 in somatotropinomas which might have an important clinical implication as both b-arrestins are able to bind to GPCR [9]. In fact, b-arrestins have been shown to affect the rate of sst2 and sst5 internalization/recycling and desensitization [26][27][28]. However, we did not find any significant correlation between b-arrestins 1 and 2 with sst2, sst5 (mRNA and protein levels) or D2 levels, which is also in contrast with a previous study showing an association between b-arrestin 1 and 2 mRNA levels with sst2 protein expression [21].
Notably, our study revealed that no correlation exists between barrestins 1 or 2 mRNA levels and the response to first-generation SRL treatment in somatotropinomas. In contrast, previous studies by Gatto et al. found a negative correlation of b-arrestin 1 with response to acute octreotide test in vivo and of b-arrestin 1 and b-arrestin 2 with long-term treatment with first-generation SRL [21,24]. Moreover, they found a negative correlation of b-arrestin 1 with octreotide treatment in vitro [24]. In fact, it has been suggested in this report that lower levels of b-arrestins could lead to a greater number of biologically active receptors on the cell membrane and consequently better response to treatment. Many reasons may explain the discrepancy between this previous study and our present study. Specifically, one of the reasons could be the different criteria to define response to SRL. Gatto et al. [21] defined biochemical control with GH levels lower than 2.5 lg/l and age-matched IGF-I levels below the ULNR, while we used a different cut-off point for GH (lower than 1.0 lg/l). Another possible reason is that Gatto et al. [21] considered at least 3 months as the minimum time of treatment to define response to SRL, while we believe that a longer duration of treatment is necessary to better define response to treatment and that is the reason why we defined treatment outcome after 6 months of therapy. In addition, another reason for such difference could be based on the methodology used, in that the qPCR analyses used in our study included three housekeeping genes as internal controls, while Gatto et al. [21] used only one housekeeping gene which might alter the relative expression levels of the transcripts analysed across the samples as no unequivocal single reference gene (with proven invariable expression between cells/tissue samples) has been identified yet [29]. Therefore, as the best alternative, the mean expression of multiple housekeeping genes is recommended to be used for normalization of tumoral samples [30]. Furthermore, it should be mentioned that the absence of barrestins correlation with the long-term response to first-generation SRL in patients with somatotropinomas observed in the present study is in agreement with the clinical data showing an absence of tachyphylaxis during this treatment [4]. However, single-centre studies in acromegaly, as ours, usually have a small number of patients and, therefore, a type II statistical error cannot be excluded. In addition, Gatto et al. [21] found that sst2/b-arrestin 1 and sst2/b-arrestin 2 ratios seem to be a better predictor in response to first-generation SRL compared to sst2 mRNA analysis alone. However, in our case, the relationship did not prove to be a better predictor compared to sst2 alone (data not shown). Therefore, as these are the only two studies in the literature evaluating the influence of the b-arrestins in the response to long-term first-generation SRL treatment, additional studies are necessary to clarify this issue.
Previous studies had shown involvement of both b-arrestins at cell cycle regulation, cell migration and apoptotic signalling in other tumours [12], and changes in b-arrestins expression were associated with more aggressive cancer phenotypes [12][13][14]. To the best of our knowledge, this is the first study that evaluated the relation between b-arrestin 1 or 2 mRNA expression and tumour invasiveness in somatotropinomas. However, different from what has been found in other tumour types [12,31,32], we did not find such a relation in somatotropinomas.

Conclusion
We did not observe a correlation of b-arrestins with sst2, sst5, DR2 expression nor a relation of its expression with the response to firstgeneration SRL long-term treatment or tumour invasiveness in this subset of patients with acromegaly. Considering the divergence of our data with previous data in the literature, further studies are necessary to clarify whether b-arrestins have a role in predicting the response to first-generation SRL treatment in patients with acromegaly.
Instituto de Salud Carlos III, co-funded by European Union (ERDF/ESF, 'Investing in your future' PI16/00264) and CIBERobn (to RML). CIBER is an initiative of Instituto de Salud Carlos III, Ministerio de Sanidad, Servicios Sociales e Igualdad, Spain.

Disclosure
MRG has received unrestricted research grants and lecture fees from Novartis, Ipsen and Pfizer, has participated on advisory boards of Novartis and Ionis and is PI in clinical trials by Novartis and Ipsen. The other authors have nothing to disclosure.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Table S1 b-arrestin 1, b-arrestin 2, sst2, sst5, D2 mRNA levels in individual somatotropinomas estimated mRNA copy number corrected by a normalization factor (NF) derived from the expression of three housekeeping genes (glyceraldehyde-3-phosphate dehydrogenase -GAPDH, b-actin -ACTB and hypoxanthine-guanine phosphoribosyl transferase -HPRT) together with sst2 and sst5 protein evaluation (as IRS)